Received: 29 May 2019 | Revised: 26 July 2019 | Accepted: 7 August 2019 DOI: 10.1111/mec.15220 ORIGINAL ARTICLE Host plant associations and geography interact to shape diversification in a specialist insect herbivore Amanda L. Driscoe1 | Chris C. Nice1 | Robert W. Busbee1 | Glen R. Hood2 | Scott P. Egan3 | James R. Ott1 1Population and Conservation Biology Program, Department of Biology, Texas Abstract State University, San Marcos, Texas Disentangling the processes underlying geographic and environmental patterns of 2 Department of Biological Sciences, Wayne biodiversity challenges biologists as such patterns emerge from eco‐evolutionary State University, Detroit, Michigan 3Department of Biosciences, Rice University, processes confounded by spatial autocorrelation among sample units. The herbivo‐ Houston, Texas rous insect, Belonocnema treatae (Hymenoptera: Cynipidae), exhibits regional special‐ Correspondence ization on three plant species whose geographic distributions range from sympatry James R. Ott, Department of Biology, Texas through allopatry across the southern United States. Using range‐wide sampling State University, San Marcos, TX 78666. Email: [email protected] spanning the geographic ranges of the three host plants and genotyping‐by‐se‐ quencing of 1,217 individuals, we tested whether this insect herbivore exhibited host Present address Amanda L. Driscoe, Department of plant‐associated genomic differentiation while controlling for spatial autocorrelation Biological Sciences, University of Notre among the 58 sample sites. Population genomic structure based on 40,699 SNPs was Dame, Notre Dame, Indiana evaluated using the hierarchical Bayesian model ENTROPY to assign individuals to ge‐ Funding information netic clusters and estimate admixture proportions. To control for spatial autocorrela‐ Southwestern Association of Naturalists; tion, distance‐based Moran's eigenvector mapping was used to construct regression Texas State University; American Museum of Natural History variables summarizing spatial structure inherent among sample sites. Distance‐based redundancy analysis (dbRDA) incorporating the spatial variables was then applied to partition host plant‐associated differentiation (HAD) from spatial autocorrelation. By combining ENTROPY and dbRDA to analyse SNP data, we unveiled a complex mosaic of highly structured differentiation within and among gall‐former populations finding evidence that geography, HAD and spatial autocorrelation all play significant roles in explaining patterns of genomic differentiation in B. treatae. While dbRDA confirmed host association as a significant predictor of patterns of genomic variation, spatial autocorrelation among sites explained the largest proportion of variation. Our results demonstrate the value of combining dbRDA with hierarchical structural analyses to partition spatial/environmental patterns of genomic variation. KEYWORDS Cynipidae, genotyping‐by‐sequencing, host‐associated differentiation, insect–plant interactions, redundancy analysis, spatial autocorrelation Molecular Ecology. 2019;28:4197–4211. wileyonlinelibrary.com/journal/mec © 2019 John Wiley & Sons Ltd | 4197 4198 | DRISCOE ET AL. 1 | INTRODUCTION be understood to determine whether the lineage sorting occurred pre‐ or post‐host plant shifting (Forbes et al., 2017). However, ge‐ Understanding how biodiversity arises involves identifying the pro‐ nomic differentiation among populations occupying alternate hosts cesses that inhibit gene flow and reproductively isolate diverging provides evidence indicative of HAD. For example, patterns of ge‐ lineages (Bush, 1969; Coyne & Orr, 2004; Mayr, 1942; Rundle & nomic variation that consistently follow host plant associations sup‐ Nosil, 2005). Geography has received much attention in studies of port the hypothesis of ecologically diverging lineages in Rhagoletis divergence as spatial variation in biotic and abiotic interactions dif‐ fruit flies (Feder et al., 2005), Eurosta gall flies (Waring, Abrahamson, ferentially affects species ecology across their ranges and influences & Howard, 1990) and Acyrthosiphon pea aphids (Peccoud, Ollivier, the opportunity for gene flow and genetic drift between populations Plantegenest, & Simon, 2009), among others. (Coyne & Orr, 2004). These fundamental drivers of divergence, ecol‐ Herbivorous insects represent a diverse range of taxa with ogy and geography, particularly in the context of limited dispersal, their extensive diversity stemming from adaptive radiation (Mitter, predict geographically based differentiation among populations for Farrell, & Wiegmann, 1988; Schluter, 2001; Winkler et al., 2018). traits undergoing selection in different environments (Lenormand, Adaptation to new host plant species can promote ecological diver‐ 2002; Slatkin, 1993; Wright, 1943). Predictions can be drawn from gence (Funk, Filchak, & Feder, 2002) that manifests as differences different isolation scenarios involving reproductive isolation (RI), iso‐ in habitat preference, temporal isolation and sexual isolation among lation by distance (IBD), isolation by environment (IBE) or isolation insect populations occupying alternative host plants (Hood, Zhang, by adaptation (IBA). Under conditions where RI is linked to traits un‐ Hu, Ott, & Egan, 2019; Rundle & Nosil, 2005; Servedio, 2016). Given dergoing selection, isolation is predicted to accompany population the intimate linkage between insect herbivores and their host plants, differentiation (Orr & Smith, 1998). Allopatric differentiation can be regional host plant specialization (Jaenike, 1990) can arise allopatri‐ viewed as a “null model” and is predicted in the absence of selection, cally if the geographic range of insect herbivores mirrors the geo‐ where populations are separated by physical distance (or geographic graphic range expansion and (or) contraction of the host species barriers) that exceeds (or impede) dispersal capabilities thereby re‐ (Hunter & Price, 1992; Underwood & Rausher, 2000). The phylo‐ ducing gene flow (Coyne & Orr, 2004; Jenkins et al., 2010). In con‐ geographic history of host plants can also impose a biogeographic trast, IBE/IBA is predicted for species distributed across multiple history upon their specialized phytophagous insects (Althoff, 2008; environments, regardless of spatial scales, when the selection–mi‐ Berlocher & Feder, 2002; Thompson, 2005). Once geographically gration balance favours adaptation to alternative environments separated, herbivore populations may diverge via the neutral process (Funk, Egan, & Nosil, 2011; Shafer & Wolf, 2013). For widespread of genetic drift, which can result in nonecological RI (Coyne & Orr, species that throughout their range occupy alternate environments, 2004), and (or) experience differing abiotic and biotic environments which may occur in sympatry, parapatry and/or allopatry, untangling that can further alter the trajectory of the arms race between host the relative contributions of drivers of differentiation represents plant defence and herbivore counter‐defence, potentially contribut‐ a major challenge (Crispo, Bentzen, Reznick, Kinnison, & Hendry, ing to ecologically based RI among diverging populations (Futuyma & 2006; Thorpe, Surget‐Groba, & Johansson, 2008; Wang, Glor, & Agrawal, 2009). As well, incompatible alleles may evolve in different Losos, 2013). Moreover, the widespread sampling required for populations adapting to the same selective regime, resulting in RI be‐ such studies introduces structuring of patterns of genetic variation tween populations on the same host trees in geographically isolated among sample sites because of spatial autocorrelation that must be populations (Nosil & Flaxman, 2011). partitioned in order to assess the relative contributions of eco‐evo‐ The extent of historical RI can be assessed by examining patterns lutionary processes (Legendre & Fortin, 1989; Shafer & Wolf, 2013). of genome‐wide differentiation (Gompert et al., 2014; Mandeville, Here, we aim to untangle the effects of geography and envi‐ Parchman, McDonald, & Buerkle, 2015; Parchman, Buerkle, Soria‐ ronment in structuring patterns of genetic differentiation among Carrasco, & Benkman, 2016), which can illustrate population‐level populations of a widely distributed host plant‐specific insect her‐ substructuring due to either adaptive divergence and/or dispersal bivore while controlling for spatial autocorrelation among sample limitations coupled with genetic drift (Coyne & Orr, 2004; Ehrlich sites. For specialized herbivorous insects, host‐associated differ‐ & Raven, 1969; Wright, 1942, 1943), or cases of selection unrelated entiation (HAD) represents a form of IBE where the environment to adaptive divergence (i.e. selective sweeps in isolated populations) driving differentiation is the host plant (Antwi, Sword, & Medina, notwithstanding (Marsden et al., 2016). For example, genome‐wide 2015; Berlocher & Feder, 2002; Stireman, Nason, & Heard, 2005). variation among single nucleotide polymorphisms (SNPs) can indi‐ The extent to which host plant specialization affects the evolution‐ cate levels of admixture that vary along a geographic gradient (i.e. ary history of insect herbivores can vary from simple correlations IBD) or by ecological boundaries (i.e. IBE), giving insight into the with genetic differences among populations to the initiation of mechanism restricting gene flow (Lee & Mitchell‐Olds, 2011; Wang speciation (Forbes et al., 2017). Sensu stricto, patterns of HAD are et al., 2013). Here, we employ range‐wide sampling